Agenlus
Show HN: Run RL agents in the browser with WebGPU
Last verified:
What is Agenlus?
Agenlus is a browser-based reinforcement learning (RL) training platform that allows users to train RL agents directly in their web browser without any installation required. The platform leverages WebGPU acceleration to enable GPU-powered training entirely client-side, eliminating the need for expensive hardware or GPU bills. Users can train agents on classic RL environments like CartPole, MountainCar, and battle environments, then share their trained models through seamless HuggingFace integration.
Key features include a global leaderboard where users can pit their trained agents against others to compete and compare performance, the ability to push trained models directly to HuggingFace for standardized sharing in the RL ecosystem, and a solo-developed platform that launched recently on Product Hunt. The platform is designed to make RL accessible to everyone by removing compute barriers that typically lock interesting environments behind serious hardware requirements.
Agenlus is specifically built for reinforcement learning researchers, ML practitioners, students learning RL, developers experimenting with custom RL environments, and the broader RL community who want to train, share, and battle agents without infrastructure overhead. The platform aims to create compounding knowledge in the RL ecosystem similar to how HuggingFace enabled compounding progress in NLP and computer vision.
The platform addresses a critical gap in RL where great custom environments have no easy way for others to use, train on, or compete on them. By providing a standardized, accessible training platform with model sharing capabilities, Agenlus enables the RL community to compound knowledge and build on each other's work more effectively.
Agenlus pricing
Pricing model: Freemium
Free tier available - the platform is free to use with no credit card required. Users can train RL agents, share models via HuggingFace, and compete on the leaderboard at no cost. No paid plans or subscription tiers are mentioned on the website.
Agenlus pros
- No installation required - runs entirely in browser
- WebGPU-accelerated training for GPU performance
- No GPU bills or expensive hardware needed
- HuggingFace integration for seamless model sharing
- Global leaderboard to battle agents against others
- Train on classic environments like CartPole and MountainCar
- Standardized RL model sharing across the community
- Accessible to beginners without RL infrastructure knowledge
- Solo-developed platform with active community feedback
- Launch on Product Hunt with community validation
- Enables custom environment training and competition
- Client-side training protects privacy
- Free to use with no credit card required
- Fast setup - just open browser to start training
- Compounding knowledge ecosystem for RL
Agenlus cons
- Limited to browser-based training capabilities
- Requires WebGPU-supporting browser and hardware
- Relatively new platform with limited environment selection
- Solo-developed may limit feature development speed
- No mentions of advanced RL algorithms beyond basics
- Leaderboard competition may be limited by user base size
- No documented API for programmatic access
- HuggingFace integration may have rate limits
- No mobile app or offline training support
- Limited documentation for custom environment development
Frequently asked questions about Agenlus
What is Agenlus?
Agenlus is a browser-based reinforcement learning training platform where you can train RL agents directly in your browser without any installation. It uses WebGPU for acceleration, allows you to share trained agents via HuggingFace integration, and lets you battle agents on a global leaderboard. No install, no GPU bill - just open your browser.
Do I need to install anything to use Agenlus?
No installation is required. Agenlus runs entirely in your web browser with WebGPU acceleration. You simply open the website and can start training RL agents immediately without any local setup or software installation.
What RL environments are available on Agenlus?
The platform currently supports classic RL environments including CartPole, MountainCar, and battle environments. The creator has expressed interest in feedback from the RL community about what environments they would actually want to train on, suggesting more environments may be added.
How does the HuggingFace integration work?
After training an agent in the browser, you can push your trained model directly to HuggingFace. The flow is: train in browser → push to HuggingFace → others can load your model and battle it on the leaderboard. This provides standardized sharing for RL models similar to how HuggingFace works for NLP and CV.
Do I need a GPU to use Agenlus?
No physical GPU is required. Agenlus uses WebGPU acceleration that runs in your browser, utilizing your browser's GPU capabilities without requiring dedicated GPU hardware or incurring GPU bills.
Is Agenlus free to use?
Yes, Agenlus is free to use. There is no mention of paid tiers or subscription costs on the platform. Users can train agents, share models, and compete on the leaderboard without any cost.
How does the global leaderboard work?
The global leaderboard allows users to pit their trained RL agents against others. After training an agent and optionally sharing it via HuggingFace, you can battle it against other users' agents to compare performance and rank on the leaderboard.
Who built Agenlus?
Agenlus was built solo by Youngseong Kim, who launched the platform in May 2026. The creator launched on Product Hunt and is actively seeking feedback from the RL community about what environments would be most useful.
Can I create and train on my own custom environments?
The platform currently focuses on classic environments like CartPole, MountainCar, and battle environments. The creator is actively asking the RL community what environments they would want to train on, suggesting custom environment support may be考虑ed for future development.
What is the goal of Agenlus?
The goal is to make RL accessible by removing compute barriers and create compounding knowledge in the RL ecosystem. Similar to how HuggingFace made NLP and CV ecosystems compound on each other, Agenlus is built so RL agents and environments compound on each other, enabling the community to build on each other's work more effectively.